Dr. Anand Subramoney

RUB
INI
Alumni
Dr. Anand Subramoney

anand.subramoney@rhul.ac.uk

Theory of Neural Systems

Scalable Machine Learning, Theory of Neural Systems

https://anandsubramoney.com

My current research interests are:

scalable and energy efficient machine learning
understanding and developing algorithms for continual learning
creating algorithms uniquely suited for recurrent neural networks

A lot of my work derives inspiration from neuroscience and biology in the quest to build a better and more general artificial intelligence.

Selected Publications

2023

Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training
Grappolini, E. W., & Subramoney, A.
In International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA ACM

@inproceedings{GrappoliniSubramoney2023, author = {Grappolini, Edoardo W. and Subramoney, Anand}, title = {Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training}, booktitle = {International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA}, publisher = {ACM}, month = {June}, year = {2023}, }

Grappolini, E. W., & Subramoney, A.. (2023). Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training. In International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA. ACM.
Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time
Subramoney, A., Nazeer, K. K., Schöne, M., Mayr, C., & Kappel, D.
In International Conference on Learning Representations

Openreview link arXiv version

@inproceedings{SubramoneyNazeerSchöneEtAl2023, author = {Subramoney, Anand and Nazeer, Khaleelulla Khan and Schöne, Mark and Mayr, Christian and Kappel, David}, title = {Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time}, booktitle = {International Conference on Learning Representations}, month = {May}, year = {2023}, }

Subramoney, A., Nazeer, K. K., Schöne, M., Mayr, C., & Kappel, D.. (2023). Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time. In International Conference on Learning Representations. Retrieved from https://openreview.net/forum?id=lJdOlWg8td
Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity
Anand Subramoney,
In ICLR 2023 Workshop on Sparse Neural Networks

arXiv

@inproceedings{Anand Subramoney2023, author = {Anand Subramoney}, title = {Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity}, booktitle = {ICLR 2023 Workshop on Sparse Neural Networks}, month = {March}, year = {2023}, doi = {10.48550/arXiv.2303.05641}, }

Anand Subramoney,. (2023). Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity. In ICLR 2023 Workshop on Sparse Neural Networks. http://doi.org/10.48550/arXiv.2303.05641

2026

Probing Length Generalization in Mamba via Image Reconstruction
Rathjens, J., Schiewer, R., Wiskott, L., & Subramoney, A.

@misc{RathjensSchiewerWiskottEtAl2026, author = {Rathjens, Jan and Schiewer, Robin and Wiskott, Laurenz and Subramoney, Anand}, title = {Probing Length Generalization in Mamba via Image Reconstruction}, year = {2026}, }

Rathjens, J., Schiewer, R., Wiskott, L., & Subramoney, A.. (2026). Probing Length Generalization in Mamba via Image Reconstruction. Retrieved from https://arxiv.org/abs/2603.12499

2024

Exploring the limits of hierarchical world models in reinforcement learning
Schiewer, R., Subramoney, A., & Wiskott, L.
Scientific Reports, 14(1)

Link to paper

@article{SchiewerSubramoneyWiskott2024, author = {Schiewer, Robin and Subramoney, Anand and Wiskott, Laurenz}, title = {Exploring the limits of hierarchical world models in reinforcement learning}, journal = {Scientific Reports}, volume = {14}, number = {1}, month = {November}, year = {2024}, doi = {10.1038/s41598-024-76719-w}, }

Schiewer, R., Subramoney, A., & Wiskott, L.. (2024). Exploring the limits of hierarchical world models in reinforcement learning. Scientific Reports, 14(1). http://doi.org/10.1038/s41598-024-76719-w

2023

Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training
Grappolini, E. W., & Subramoney, A.
In International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA ACM

@inproceedings{GrappoliniSubramoney2023, author = {Grappolini, Edoardo W. and Subramoney, Anand}, title = {Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training}, booktitle = {International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA}, publisher = {ACM}, month = {June}, year = {2023}, }

Grappolini, E. W., & Subramoney, A.. (2023). Beyond Weights: Deep learning in Spiking Neural Networks with pure synaptic-delay training. In International Conference on Neuromorphic Systems (ICONS ′23), Santa Fe, NM, USA. ACM.
Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time
Subramoney, A., Nazeer, K. K., Schöne, M., Mayr, C., & Kappel, D.
In International Conference on Learning Representations

Openreview link arXiv version

@inproceedings{SubramoneyNazeerSchöneEtAl2023, author = {Subramoney, Anand and Nazeer, Khaleelulla Khan and Schöne, Mark and Mayr, Christian and Kappel, David}, title = {Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time}, booktitle = {International Conference on Learning Representations}, month = {May}, year = {2023}, }

Subramoney, A., Nazeer, K. K., Schöne, M., Mayr, C., & Kappel, D.. (2023). Efficient Recurrent Architectures through Activity Sparsity and Sparse Back-Propagation through Time. In International Conference on Learning Representations. Retrieved from https://openreview.net/forum?id=lJdOlWg8td
Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity
Anand Subramoney,
In ICLR 2023 Workshop on Sparse Neural Networks

arXiv

@inproceedings{Anand Subramoney2023, author = {Anand Subramoney}, title = {Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity}, booktitle = {ICLR 2023 Workshop on Sparse Neural Networks}, month = {March}, year = {2023}, doi = {10.48550/arXiv.2303.05641}, }

Anand Subramoney,. (2023). Efficient Real Time Recurrent Learning through Combined Activity and Parameter Sparsity. In ICLR 2023 Workshop on Sparse Neural Networks. http://doi.org/10.48550/arXiv.2303.05641

2022

Exploring parameter and hyper-parameter spaces of neuroscience models on high performance computers with Learning to Learn
Yegenoglu, A., Subramoney, A., Hater, T., Jimenez-Romero, C., Klijn, W., Pérez Martín, A., et al.
Frontiers in Computational Neuroscience, 46

link

@article{YegenogluSubramoneyHaterEtAl2022, author = {Yegenoglu, Alper and Subramoney, Anand and Hater, Thorsten and Jimenez-Romero, Cristian and Klijn, Wouter and Pérez Martín, Aarón and van der Vlag, Michiel and Herty, Michael and Morrison, Abigail and Diaz Pier, Sandra}, title = {Exploring parameter and hyper-parameter spaces of neuroscience models on high performance computers with Learning to Learn}, journal = {Frontiers in Computational Neuroscience}, pages = {46}, year = {2022}, }

Yegenoglu, A., Subramoney, A., Hater, T., Jimenez-Romero, C., Klijn, W., Pérez Martín, A., et al. (2022). Exploring parameter and hyper-parameter spaces of neuroscience models on high performance computers with Learning to Learn. Frontiers in Computational Neuroscience, 46. Retrieved from https://doi.org/10.3389/fncom.2022.885207

Winter Term 2022/2023

Seminars

Topics in Deep Learning for Sequence Processing

Winter Term 2021/2022

Seminars

Topics in Deep Learning for Sequence Processing

Summer Term 2022

Lectures

Introduction to Artificial Intelligence

Summer Term 2021

Lab courses

Introduction to Python

2022

Memory Modules for Deep Learning
Hark, N.
Master’s thesis, Institute of Neural Computation, Ruhr University Bochum, Bochum, Germany

@mastersthesis{Hark2022, author = {Hark, Niklas}, title = {Memory Modules for Deep Learning}, school = {Institute of Neural Computation, Ruhr University Bochum}, address = {Bochum, Germany}, month = {May}, year = {2022}, }

Hark, N. (2022, May). Memory Modules for Deep Learning. Master’s thesis, Institute of Neural Computation, Ruhr University Bochum, Bochum, Germany.

2025

Probing Length Generalization in Mamba via Image Reconstruction

We explore Mamba's length generalization capabilities through a toy vision task in which Mamba is trained to reconstruct images from patch sequences.

The Institut für Neuroinformatik (INI) is a research unit of the Faculties of Computer Science and Medicine at the Ruhr-Universität Bochum. Its scientific goal is to understand the fundamental principles through which organisms generate behavior and cognition while linked to their environments through sensory and effector systems. Inspired by our insights into such natural cognitive systems, we seek new solutions to problems of information processing in artificial cognitive systems. We draw from a variety of disciplines that include experimental psychology and neurophysiology as well as machine learning, neural artificial intelligence, computer vision, and robotics.

Contact

Universitätsstr. 150, Building NB, Room 3/32
D-44801 Bochum, Germany

Tel: (+49) 234 32-28967
Fax: (+49) 234 32-14210

sekretariat@ini.rub.de
www.ini.rub.de